Video frames text detection through Bayesian classification and boundary growing method

Abstract

Text detection in natural scene images is important for content-based image analysis. It consists of different component analysis for text detection video frames. Multioriented text detection in video frames is one of the text detection in videos frames. Multi-oriented text detection is not easy as detection of captions or graphics which is usually appears in horizontal direction and has high contrast compared to its background. Multi-oriented text detection generally refers to scene text that makes text detection more challenging. Therefore conventional text detection may not give good results for multi-oriented video frames text detection. Hence, in this paper, we present a new enhancement method that includes the product of Laplacian and Sobel operations to enhance text pixels in videos. To classify true text pixels, we propose a Bayesian classifier without assuming a priori probability about the input frame but estimating it based on three probable matrices. Three different ways of clustering are performed on the output of the enhancement method to obtain the three probable matrices. Text candidates are obtained by intersecting the output of the Bayesian classifier with the Canny edge map of the input frame. The robustness of the methods are analyzed our own datasets under different video frames. Finally, it coarse-to-fine detection locates text regions efficiently.